HIV testing, behavior change, and the transition to adulthood in Malawi

HIV Testing, Behavior Change, and the Transition to Adulthood in MalawiAuthor(s): Kathleen Beegle /Michelle Poulin /Gil ShapiraSource: Economic Development and Cultural Change, Vol. 63, No. 4 (July 2015), pp. 665-684Published by: The University of Chicago PressStable URL: http://www.jstor.org/stable/10.1086/681232 .

Accessed: 10/06/2015 09:09

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access toEconomic Development and Cultural Change.

http://www.jstor.org

This content downloaded from 138.220.40.146 on Wed, 10 Jun 2015 09:09:31 AMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=ucpress

http://www.jstor.org/stable/10.1086/681232?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


HIV Testing, Behavior Change, and the
Transition to Adulthood in Malawi
kathleen beegle, michelle poulin, and gil shapira
World Bank
For young women and men, decisions about school attendance, sexual and
marital partnerships, and fertility shape one’s life course and have long-termimplications on well-being. Such decisions typically involve a trade-off be-tween short-term and long-term utility in the face of much uncertainty. TheAIDS epidemic in much of sub-Saharan Africa further complicates the frame-work within which these kinds of trade-offs are considered. First, these deci-sions directly affect exposure to risk of HIV infection. Young people are wellaware that some avenues toward adulthood ðe.g., leave school, marry soonafter, have a child soon afterÞ lead to different risk of exposure to HIV infectionthan others ðPoulin 2007; Clark, Poulin, and Kohler 2009Þ. Second, youngpeople may face these pivotal decisions with uncertainty about their currentHIV status and expected life horizon.Throughout sub-Saharan Africa, young people are coming of age at a time
when AIDS policy emphasizes HIV counseling and testing. International andlocal public health communities view HIV testing and the counseling that ac-companies it as the gateway to treatment. But even for those who test nega-tive, it is also hoped that testing will result in preventive behaviors that slowthe spread of the epidemic.1 There has been a rapid expansion of voluntarycounseling and testing ðVCTÞ coverage in Malawi, achieved through outreachand mobilization initiatives such as Malawi’s annual “Testing Week” and anincreased supply of such services ðAngotti 2010; Angotti, Dionne, andGaydosh 2011Þ. Data from the 2004Malawi Demographic andHealth SurveyðMDHSÞ show that only 15% of men had been tested during the previous

The views expressed here do not necessarily reflect those of the World Bank or its member countries.
The authors are grateful for comments from anonymous reviewers, Sarah Baird, Erick Gong, RachelHeath, Jennifer Johnson-Hanks, Susan Watkins, and participants at the 2012 Mini-Conference onMarriage and HIV in San Francisco, CA, and at the PopPov Conference on Population, ReproductiveHealth, and Economic Development in Accra, Ghana. Contact the corresponding author, KathleenBeegle, at [email protected] For example, the government of Malawi’s ð2003Þ National HIV/AIDS Policy report stronglypromotes testing as a general prevention policy through a reduction in risky behaviors, as well asthrough improvement in access to treatment and mother-to-child transmission prevention.
Electronically published April 8, 2015© 2015 by The University of Chicago. All rights reserved. 0013-0079/2015/6304-0001$10.00



year, whereas data from the from MDHS 2010 show that 51% of men had ever

666 E C O N O M I C D E V E L O P M E N T A N D C U L T U R A L C H A N G E

been tested and 31% of men reported having been tested in the past 12 monthsðNSO and ICF Macro 2011Þ. Women were more likely to have been testedin the past 12 months ð72% testedÞ. This partly reflects the introduction oftesting through ante-natal surveillance sites in the early years of the epidemic.In this article we use the Marriage Transitions in Malawi ðMTMÞ data set, a

panel study of initially never-married young women and men in central Malawi,to evaluate how, if at all, VCT influences behaviors of young people. Thesurvey introduced a testing opportunity for a random set of respondents dur-ing the second year of the 3-year panel. We examine behaviors closely linkedwith the transition to adulthood and HIV risk. In particular, we estimate theintent-to-treat effect of the VCT intervention on school attendance, marriage,fertility, and reported sexual behavior in the year after the test.Testing may alter these behaviors in several ways. Consider its potential ef-

fect on sexual behavior. If a man discovers he is negative, for example, he couldcommit to maintaining his status by carefully choosing partners, by choosingmonogamy, by using condoms, or perhaps by seeking circumcision. A positiveresult might lead to a decision to engage in risky sexual behavior as incentivesto protect against infection might no longer exist. But, this same man could beconcerned about infecting others and therefore opt for safer behaviors.Taking a broader view, removing uncertainty about one’s status may affect

expectations of young people about their life expectancy as well as opportu-nities that will be available to them in the future. In turn, it may alter presentor near-term decisions with long-term implications, such as investments in hu-man capital through staying in school, selection of partners, and timing of mar-riage and fertility. Beliefs about one’s own HIV status can also translate intobeliefs about the survival of ðyet to be bornÞ children, since the virus can betransmitted from mothers to their newborns. Therefore, learning one’s statusmight affect the desired number and timing of births. Trinitapoli and Yeatmanð2011Þ, for example, find that compared to those who express certainty that theyare HIV negative, those who are uncertain about their HIV status are likely todesire accelerated childbearing.These mechanisms presume that individuals update their beliefs about their

own status after learning their test result. Behavioral responses to a test there-fore may depend on the extent to which one’s beliefs about one’s infectionstatus changed. For example, an unmarried, 18-year-old woman who believesher likelihood of current infection to be low may be less likely to change herbehavior after a negative test result compared to a young woman who believedher likelihood to be high but learns she is not infected.



VCT has the potential to affect behaviors beyond the impact of a revealed

Beegle, Poulin, and Shapira 667

test result. AVCT intervention includes pre- and posttest counseling by trainedand certified counselors. During these sessions, counselors not only provide in-formation about how to avoid infection but also encourage deliberate decisionmaking with respect to sexual and reproductive health ðAngotti 2010Þ. Al-though our data show that almost all respondents were aware of correct waysto avoid infection ðsee also Watkins 2004Þ, the interactions with the counsel-ors could have motivated them to behave in certain ways, especially those thatwould reduce the risk of contracting HIV.We begin our analysis by examining the overall effect of VCT on selected

behaviors of young men and women. We then proceed to examine the effectsof VCT by prior beliefs, employing empirical strategies similar to Boozer andPhilipson ð2000Þ and Gong ð2015Þ. The results of both analyses show a neg-ligible effect of VCTon the considered behaviors, with modest effects for mensuggesting a slower transition to adulthood. We then examine heterogeneouseffects of VCT by household wealth. We choose to focus on wealth becauseour data, as well as other studies, show a very strong correlation between wealthand the behaviors that our study focuses on. Wealth affects not only the set ofpossible choices available to young adults, but it can also influence expectationsabout opportunities and well-being in the future. The results of this analysisindicate that the VCToffer is associated with lower marriage rates and initiationof fertility among poorer young women who report some likelihood of beinginfected and wealthier young men. This is ex post analysis that was not includedin the experimental design. Therefore, while informative, the results shouldbe cautiously interpreted.

Evidence on TestingThere are a handful of studies purposively designed to examine the impact oftesting on risky sexual behaviors.2 Delavande and Kohler ð2012Þ, in investi-gating this phenomenon with a sample of adults in Malawi, exploit a randomassignment in vouchers for cash to be redeemed upon retrieval of one’s testresult from temporary VCTsites. They examine the effect of testing on a rangeof risky behaviors and show that selectivity into testing is important for in-ferring the impact of testing. Learning one is HIV-positive results in fewerpartners and more condom use up to 2 years later. Using the same voucherexperiment as Delavande and Kohler, Thornton ð2008Þ finds that individuals

2 Gersovitz ð2011Þ discusses studies that explore the implications of testing with nonrandom testing.
See also Potts et al. ð2008Þ.


who receive a positive test result are more likely to purchase condoms 2 months


later. However, she does not find a significant effect for those who learned anegative test result. She also does not find that beliefs about one’s status beforetesting matter in terms of impact of testing.Other studies have also considered the role of beliefs about own infection

and the updating of beliefs after learning a test result. Using data from Nai-robi, Kenya, and Dar es Salaam, Tanzania, Gong ð2015Þ finds that individualswho learn that they are positive ðand did not think so before that testÞ are morelikely to engage in risky sexual behavior after being tested—what he describes asan “unintended consequence of testing” ð33Þ.3 In San Francisco, a very differ-ent setting but among the earliest empirical studies on the topic, Boozer andPhilipson ð2000Þ find that among unmarried individuals, testing induces achange in behavior only if the tested person was surprised by the test result.Like the Gong study, this finding stresses the importance of prior states ofuncertainty, or prior beliefs, in learning the results of a test.Baird et al. ð2014Þ report results of an experiment similar to the one

described in this article.4 Their sample was of adolescent girls in one districtin southern Malawi who were randomly offered VCT in 2009. The random-ization was across communities ð52 getting VCT and 36 with no VCTÞ. Thesample was reinterviewed 10 months later. Baird and colleagues found thatlearning a positive test result led to an increase in the likelihood of contractingHerpes Simplex Virus, with a higher likelihood of contraction for those sur-prised by the test result. Among those who tested negative, achievement testscores were improved, a finding they interpret to be that those with longerperceived life horizons have greater incentive to invest in human capital.In this study, we evaluate the effect of testing on outcomes for both young

women and men. We examine the effect of VCTon sexual behavior but alsoon a range of other interrelated outcomes specific to the transition from ad-olescence to adulthood. Focusing on this period during the life cycle is ofspecific policy relevance because of the emphasis on youth as a targetedHIV at-risk population throughout sub-Saharan Africa ðPoulin, Dovel, andWatkins 2014Þ.3 The characteristics of Gong’s sample are different from those of the Malawi sample used byDelavande and Kohler and Thornton in several dimensions that could potentially explain the
opposite results. Individuals in his sample are less likely to be married and reside in urban areas withhigher HIV prevalence than the largely rural areas where the Malawi data were collected. Gong’ssample consists of people who were seeking HIV-related services, not a random sample of thepopulation. Two-thirds of the baseline sample attrite by follow-up 6 months later.4 Although the age ranges are almost the same between Baird et al. ð2014Þ and ours ðwomen age 13–22 and 14–21, respectivelyÞ, there are notable differences. The women in their sample aremore likelyto be in school ð75% compared to 43%Þ and less likely to be married ð9% compared to 21%Þ. Oursample is closer to national statistics from the MDHS 2010. Ninety percent of young women in theBaird et al. ð2014Þ study report no chance of being infected withHIV, compared to 71% in our sample.


Setting and Data


The potential for VCT to affect behaviors related to the transition into adult-hood is important when considering the social and economic environmentsfacing young people in a poor country like Malawi. Education levels remainvery low. According to the MDHS from 2010, 16% of young men age 25–29had completed secondary school, and fewer than 5% had attended postsec-ondary school; among women in this age group, 8% had completed secondaryschool and less than 3% had attended school beyond the secondary level ðNSOand ICF Macro 2011Þ. Poverty remains high. There are few opportunities fornonfarming employment and secondary education, age at first marriage re-mains young, and nearly everyone marries at least once. Median ages at firstmarriage for respondents age 25–49 are 17.8 for women and 22.5 for men.Malawi has a generalized HIV epidemic; prevalence among the 15–49-year-

old population is estimated to be 10.6% ðNSO and ICF Macro 2011Þ with asteep age gradient. The HIV prevalence among 15–17-year-old young womenis 3.4%, while that of 18–19-year-olds is 5.7%. The prevalence is 1.3% among15–19-year-old men but increases to 4.6% among 23–24-year-olds and 6.9%for those 25–29. Prevalence rates among never-married women and men ðourstudy sample at baselineÞ are about half of the rates among those ever married.Prevalence rates are much higher in urban Malawi than in rural areas like ourstudy site.This study uses data from the MTM project, a panel survey conducted in

60 rural and semiurban communities in the Salima district of central Malawi.The project was designed to understand socioeconomic patterns of youngadults as they transition into adulthood and with an emphasis on HIV/AIDS.The connection between the two is motivated largely by the search for a spouse.This search is associated with an assessment of potential partners, leading tochanges in partners or in unprotected sex to ensure the potential spouse is fertileðClark et al. 2009; Poulin and Beegle 2014Þ. In some countries in south andeastern Africa, the rate of new infections rises during this search process ðGlynnet al. 2003; Magruder 2011Þ.The study consists of 1,183 initially never-married young women and men

and was designed to follow them into marriage.5 Three annual householdsurveys were conducted ð2007, 2008, 2009Þ, with two Partnership Interviewsurveys conducted midway between the annual surveys.6 The data contain

5 Sixty enumeration areas were randomly selected from a sample of 215 areas stratified by distance to

6 In between the three main summer rounds from 2007 to 2009, the MTM study included aninterim survey round for a two-thirds of the sample randomly selected and then interviewed with a

main trading centers. The sampling frame for respondents within the enumeration areas was strat-ified by age for men and women. More information on the sample design is available at http://www.sites.google.com/site/mtmalawiproject/mtmbackground.



detailed information on partnering behavior as well as socioeconomic condi-


tions, including asset ownership and family background. Since this age groupis highly mobile, the MTM study made additional efforts to track sample re-spondents who relocated after the baseline round in 2007 ðBeegle and Poulin2013Þ. This is especially important since marriage itself often results in movingto a new village or town. Tracking proved important for ensuring recontact rates;more than one-quarter of respondents moved during the course of the panel.For this analysis, we include respondents interviewed during all three annual

household surveys. This sample consists of 1,009 respondents, 529 women and480 men. In the 2008 round, a random sample of respondents was offeredan HIV test.7 Of the 1,009 respondents, 531 ð53%Þ were randomly offeredVCT. The test was administered by trained and certified VCT counselors in theprivacy of each respondent’s residence. The test result was offered immediately.Ninety-three percent of those offered a test consented to it; of these, only fiverespondents opted to not learn their status.8 There are only minimal differencesbetween those who refused the test and those who accepted: men with noschooling were less likely to accept the test; women from wealthier householdsweremore likely to accept the test. Bothmen andwomenwhohadbeen tested inthe past were more likely to accept the offer of a test. Of those tested, less than1% of the men ð2 out of 237Þ and 2% of the women ð4 out of 274Þ were HIVpositive.In table 1 we report sample characteristics at the 2008 round, by gender

and by whether respondents were offered an HIV test. The results reflect therandomization process, whereby the principal investigators generated a ran-dom number program in Stata for selection. The variables are well balancedacross the control and treatment groups. Variable-by-variable individual testsreported in the table cannot reject that the means are the same for the twogroups for almost all the variables.Women in the sample range in age from 14 to 21 ðby designÞ, with a mean

age of 17. Nearly all women have some schooling, and 32% have attended

modified questionnaire. These Partnership Interviews focused on significant life events over the

7 Of the 174 respondents who are not in either round 2 or 3, three-quarters were not found, and therest refused to participate. Those lost to attrition have comparable sociodemographic characteristicsas the tracked sample ðresults not reportedÞ.8 For comparison, note that Trinitapoli and Yeatman ð2011Þ had an acceptance rate of 80% for asample of young adults in southern Malawi in 2009. MDHS 2010 had testing rates of about 90% forwomen and 83% for men ð15–24Þ. Baird et al. ð2014Þ tested 98% of their sample of adolescentwomen.

6 months prior, especially on changes in sexual and partnership behavior. We do not use this roundfor this analysis for two reasons. First, the roughly 5–6 months between summer 2008 and theinterim round in 2009 may be too short for the behavior changes we examine. Second, we want toexamine results for the complete sample.



TABLE

1SA

MPLE

CHARACTE

RISTICSBYGENDERAND

TREATM

ENT

Youn

gWom

enYo

ungMen

All

ðN5

529Þ

Con

trol

ðN5

233Þ

Trea

tmen

tðN

529

6ÞDifferen

ceðt-

TestÞ

All

ðN5

480Þ

Con

trol

ðN5

245Þ

Trea

tmen

tðN

523

5ÞDifferen

ceðt-

TestÞ

Age

16.68

16.72

16.65

.07

20.34

20.47

20.21

.26

ð1.63Þ

ð1.68Þ

ð1.59Þ

ð.14Þ

ð1.83Þ

ð1.92Þ

ð1.74Þ

ð.17Þ

Scho

oling:

Non

e.02

.02

.01

.01

.03

.04

.03

.004

Prim

ary

.66

.69

.64

.06

.62

.62

.63

−.01

Seco

ndary

.32

.28

.35

−.07

.31

.29

.33

−.03

Tribe:

Che

wa

.63

.63

.63

−.01

.65

.67

.63

.04

Yao

.19

.18

.20

−.01

.19

.19

.19

.001

Ngon

i.09

.10

.09

.01

.08

.08

.08

−.004

Wea

lth:

2ndqua

rtile

.24

.27

.22

.05

.27

.26

.28

−.02

3rdqua

rtile

.26

.26

.27

−.01

.25

.24

.26

−.02

4thqua

rtile

.24

.22

.26

−.05

.22

.24

.21

.03

Marrie

d.21

.21

.20

.01

.16

.15

.17

−.02

Inscho

ol.43

.42

.44

−.02

.24

.22

.26

−.03

Sexu

ally

activ

e.44

.47

.42

.05

.59

.61

.58

.03

Ever

hadsex

.59

.62

.57

.05

.85

.88

.82

.06*

Ever

pregna

nt.15

.16

.14

.02

.16

.14

.19

−.04

Tested

before

.40

.45

.37

.09

.53

.52

.53

−.01

Reports

nolikelihoo

dof

being

infected

.71

.71

.71

.00

.92

.95

.90

.05**

Note.Outco

mes

inMarria

geTran

sitio

nsin

Malaw

i200

8roun

d.W

iththeex

ceptio

nof

age,

allcov

ariatesarebinaryindicators.Stan

darddev

iatio

nsin

paren

theses.T

reatmen

trefers

tobeing

offeredatest.C

ontrol

indicates

notest

offer.

*Significant

at10

%leve

l.**

Significant

at5%

leve

l.



secondary school. At the time of the testing offer in 2008, 43% were still


attending school. Fifty-nine percent of women reported ever having sex, and44% reported having had sex in the 12 months before the 2008 interview.Fifteen percent reported ever being pregnant and 21% were married.The average age of the men in the sample is 20, ranging from 14 to 26.9

Thirty-one percent of men have attended at least some secondary school, and62% have attended some primary school. At the time of the 2008 interview,24% of men were attending school. The lower percentage of men attendingschool reflects the older ages relative to the women. The men were also morelikely than women to have ever had sex and to report being currently sexuallyactive. Eighty-five percent reported ever having sex, and 59% reported beingsexually active. Sixteen percent of men reported ever having impregnated awoman. The same proportion of men was married. Because of the widespreadavailability of VCT in Malawi, it is not surprising that just under half of thesample had been tested before the testing offered by the MTM VCT team.Men were more likely to have been tested than women ð53% vs. 40%Þ.As discussed above, those tested, and especially when recently tested, may

respond differently to an additional test than would those tested for the firsttime in their lives. Table 2 shows the traits of men and women in 2008 bytheir prior testing status. Those with a prior testing experience were on averageolder, had higher education, and were more likely to be married, sexually ac-tive, and to have ever been pregnant ðfor womenÞ or to report having impreg-nated a woman ðfor menÞ. In each interview round, respondents reported onthe likelihood they were infected with HIV. Respondents chose one of fourcategories: no likelihood, low likelihood, medium likelihood, and high like-lihood. Table 3 shows these stated beliefs in 2008, just before the testing offer.The table reports the beliefs overall and by some specific traits. Generally thissample reported low levels of beliefs about their own infectivity; very fewreported a medium or high likelihood that they were HIV infected, so thatmost fall within the “no likelihood” or “low likelihood” categories.10 There arenotable gender differences in the reporting of low likelihood. While 28% ofyoung women assigned at least some likelihood to being infected, only 8% ofmen reported the same. Some variation also exists across social and demo-graphic groups. T-tests confirm that respondents were significantly more likely

9 Because of the later age of marriage for men, the MTM project purposefully aimed for the sample
of young men to be older than that of young women.10 This distribution likely reflects the young age of the respondents, although the same questionposed to adults in rural Malawi yielded a similarly lopsided distribution, albeit with slightly greater per-centages falling into the medium and high likelihood categories ðSmith andWatkins 2005Þ.


to assign at least some likelihood to being infected if they ever had sex, were

TABLE 2SAMPLE CHARACTERISTICS BY TESTING BEFORE 2008 ROUND

Young Women Young Men

Not TestedðN 5 316Þ

TestedðN 5 213Þ

Differenceðt-TestÞ

Not TestedðN 5 228Þ

TestedðN 5 252Þ

Differenceðt-TestÞ

Age 17.35 18.16 −.82*** 21.09 21.56 −.47***Schooling:

Primary .71 .58 .13*** .69 .56 .13***Secondary .27 .39 −.12*** .26 .36 −.10**

Tribe:Chewa .63 .63 .00 .68 .62 .06Yao .21 .17 .04 .17 .2 −.03Ngoni .08 .12 −.04 .07 .08 −.01

Wealth:2nd quartile .25 .22 .03 .28 .26 .023rd quartile .25 .29 −.04 .26 .25 .014th quartile .22 .27 −.04 .18 .26 −.08**

In school .50 .33 .17*** .25 .23 .01Married .12 .33 −.21*** .11 .19 −.08**Sexually active .33 .6 −.27*** .54 .64 −.10**Ever had sex .48 .75 −.27*** .83 .86 −.03Ever pregnant .04 .30 −.26*** .11 .22 −.11***

Note. Outcomes in Marriage Transitions in Malawi 2008 round.** Significant at 5% level.*** Significant at 1% level.


currently sexually active, or were ever pregnant. Young womenweremore likely toassign no likelihood if they were attending school and less likely to do so if theywere married. Men were more likely to assign no likelihood of being infectedif they had been tested for HIV before the interview. The analysis that followsexamines the differential results of a test offer by these self-reported likelihoods.

Estimated Causal Effects
To examine the impact of testing on subsequent behaviors, we study theintent-to-treat effect on six outcomes that capture key events experienced byyoung people in the transition from adolescence to adulthood. We focus onschool attendance, marriage, fertility, and sexual behaviors as measured in 2009,a year after the VCToffer. Specifically, the six outcome indicators used in theanalysis are ðiÞ stay in school:11 an indicator for whether a respondent is at-tending school in 2009, conditional on being in school in 2008 ðrespondents
11 Because reentry into primary or secondary school after leaving school is rarely observed in our
sample, we examine school enrollment conditional on being enrolled in the baseline, in 2007.


who report the highest grade level of secondary school as the highest grade12

TABLE 3LIKELIHOOD OF BEING HIV INFECTED IN 2008

Young Women Young Men

No Low Medium High No Low Medium High

All .71 .25 .02 .01 .92 .05 .02 .00Schooling:

Primary .72 .25 .02 .01 .92 .05 .03 .00Secondary .70 .27 .02 .01 .95 .05 .00 .00

Tribe:Chewa .73 .25 .02 .00 .92 .06 .02 .00Yao .71 .28 .00 .01 .94 .01 .02 .02

Wealth:2nd quartile .74 .26 .01 .00 .89 .09 .02 .013rd quartile .74 .23 .03 .00 .98 .02 .01 .004th quartile .67 .31 .02 .01 .90 .08 .03 .00

In school .78 .21 .00 .01 .94 .05 .01 .00Married .63 .34 .02 .01 .92 .05 .01 .01Sexually active .64 .32 .03 .01 .89 .08 .02 .01Ever had sex .65 .31 .03 .01 .91 .06 .02 .00Ever pregnant .62 .33 .05 .00 .86 .08 .04 .03Tested before .68 .28 .03 .01 .94 .03 .02 .01

Note. Share of respondents for each trait reported by the four categories to answer the question “In youropinion, what is the likelihood ðchanceÞ that you are infected with HIV/AIDS now?”


attended by 2008 are excludedÞ; ðiiÞ got married: married in 2009 condi-tional on having not being married in 2008; ðiiiÞ first pregnancy: pregnancybetween 2008 and 2009 conditional on no pregnancy before the VCT in-tervention ðfor men, this is an indicator for ever impregnating a sexual part-nerÞ; ðivÞ number of pregnancies: cumulative number of pregnancies ðfor men,it includes pregnancies for all partnersÞ; ðvÞ sexually active and not yet married:reporting having had sex in the 12 months before the 2009 interview, exclud-ing respondents who are married in 2008; and ðviÞ multiple partners: an indi-cator for reporting more than one sexual partner in the past year.Given the random allocation of testing offers and the balance between the

treatment and control groups, differences in outcomes’means between groupscan be attributed to the VCT intervention and interpreted as intent-to-treateffects. Therefore, we present our results in terms of mean-comparison tests.13

12 There are very high barriers to continuation from secondary to tertiary education in Malawi.
13 Age is highly correlated with most of the outcomes that we focus on. Therefore, although theaverage age in the treatment and control groups is overall balanced, the age composition of the groupsmight influence the results when we do the subgroup analysis or restrict the sample for some of ouroutcomes. To verify that this is not driving our results, we performed an additional analysis of the fullset of results presented in the article by regressions with age fixed effects. This analysis resulted intreatment effects that are similar to those presented here.


As discussed above, the effect of learning one’s HIV status may depend on


the individual’s beliefs about his or her own HIV status before the test offer.In a manner similar to the empirical approaches employed by Boozer andPhilipson ð2000Þ and Gong ð2015Þ, we explore how the testing effect variesby these beliefs. We analyze separately the effects on the group that assignedno likelihood to being HIV infected in 2008 from those on the group that as-signed any likelihood. We combine the low, medium, and high likelihood cat-egories because less than 3% of respondents chose the medium or high catego-ries. We observe very few men who express any likelihood of being infected,so we perform this subgroup analysis for our sample of young women only.Table 4 presents results of the main specification for young women. Each

column presents results for an estimation of one of the six outcomes of in-terest. We do not find a statistically significant effect of VCTon any of the out-comes. Moreover, the differences between the study groups are small in mag-nitude. These findings—a lack of impact of VCT—are similar to the outcomesfor the sample of women in Baird et al. ð2014; with the caveat in n. 4 on sam-ple comparabilityÞ.Table 4 also presents results for young men.14 As for the young women, we

observe little response to the test offer. All the outcomes indicate a slowertransition to adulthood among the group who received the VCToffer. How-ever, for four out the six outcomes, the difference between the groups is notstatistically significant. Men who are not married when tested are 7 percentagepoints less likely to be married a year after receiving VCT offer. Men whoreport having never impregnated a woman before getting tested are 8 per-centage points less likely to impregnate a woman in the following year. Both ofthese effects are significant at the 10% level. These findings’ lack of signifi-cance for most outcomes leads us to conclude that there are negligible effectsfor young men.There are several reasons why offering an HIV test for young people might

have little or no impact on behaviors. In settings such as Malawi where testingis now common, a new test might only provide marginal information, reflect-ing risk of exposure since the previous test and not since first sex. In addition,young people have had short durations of exposure to risk of HIV infection.In our data this is reflected in the results of the HIV tests, where only a hand-ful of respondents were found to be HIV infected. It is also reflected in the lowlevels of likelihood that respondents assigned to being infected before takingthe test. Consequently, for many of the respondents it is likely that learning

14 The sample sizes for the schooling outcomes conditional on 2008 enrollment are smaller since the
sampled men are older and, therefore, less likely overall to be in school in 2008.


test results did not provide new information that would alter behaviors. Third,

TABLE 4EFFECT OF VOLUNTARY COUNSELING AND TESTING ON OUTCOMES IN 2009

Stay inSchoola

GotMarriedb

FirstPregnancyc

No. ofPregnancies

SexuallyActived

MultiplePartnerse

Young Women

Control .624 .234 .379 .502 .402 .009Test offer .723 .191 .341 .446 .447 .024Difference −.099 .043 .038 .056 −.045 −.015

ð.065Þ ð.040Þ ð.046Þ ð.053Þ ð.049Þ ð.011ÞSample size:

Control 93 184 195 233 184 233Test offer 119 236 255 296 235 296

Young Men

Control .683 .181 .290 .434 .575 .142Test offer .731 .113 .213 .386 .521 .094Difference −.048 .068* .078* .048 .054 .047


Control 41 193 193 226 193 226Test offer 52 212 207 254 211 254

Note. Mean outcomes as measured in the 2009 round. Standard errors of the difference in means inparentheses.a Excluding respondents in the fourth year of secondary school.b Conditional on not being married in the 2008 round.c Conditional on no pregnancies by the 2008 round.d Reports having sex in the 12 months before interview, excluding respondents who report being marriedin 2008.e Reports more than one sexual partner in the 12 months before interview.* Significant at 10% level.


learning a negative test result, as most of our respondents did, provides infor-mation only about current HIV status. As these youth reside in a setting with ageneralized epidemic, they might still perceive high risk of infection associatedwith different behaviors and high uncertainty for future status.Table 5 presents the results of our second specification, interacting testing

with prior beliefs for young women. This is an ex post analysis, as the random-ization was not designed in regard to prior beliefs. We find that reported like-lihood of being HIV positive does not change the results above. For those whoassigned at least some likelihood to being infected with HIV, only one of thesix outcomes is significant at the 10% level: the testing offer resulted in a de-crease of 13 percentage points of marriage for those who were not married in2008 when they reported a likelihood of being positive. Among those whoreport no likelihood of being infected, we find that a test offer yields a positiveeffect on remaining in school in subsequent rounds ðsignificant at the 10%levelÞ. Yet we find no significant effect on fertility and sexual behaviors,



regardless of the level of prior beliefs. We conclude that there is basically no

TABLE 5EFFECT OF VOLUNTARY COUNSELING AND TESTING BY PRIOR BELIEFS, YOUNG WOMEN

Stay inSchoola

GotMarriedb

FirstPregnancyc

No. ofPregnancies

SexuallyActived

MultiplePartnerse

Likelihood:Control .706 .292 .463 .591 .542 0Test offer .643 .159 .418 .588 .516 .012Difference .063 .133* .045 .003 .026 −.012

ð.148Þ ð.079Þ ð.091Þ ð.101Þ ð.097Þ ð.013ÞNo likelihood:

Control .605 .215 .348 .464 .356 .012Test offer .747 .202 .314 .389 .422 .028Difference −.142* .013 .034 .075 −.066 −.016


Likelihood:Control 17 65 54 66 65 66Test offer 28 98 67 85 98 85

No likelihood:Control 76 65 141 166 65 166Test offer 91 98 188 211 98 211

Note. Mean outcomes as measured in the 2009 round. “Likelihood” represents whether respondentassigned any likelihood to being infected with HIV in the 2008 round. Standard errors of the difference inmeans in parentheses.a Excluding respondents in the fourth year of secondary school.b Conditional on not being married in the 2008 round.c Conditional on no pregnancies by the 2008 round.d Reports having sex in the 12 months before interview, excluding respondents who report being marriedin 2008.e Reports more than one sexual partner in the 12 months before interview.* Significant at 10% level.


impact of testing for those who reported any likelihood ðmostly lowÞ of beinginfected or for those who reported no likelihood.

Wealth EffectsWealth is closely linked with the behaviors considered in this article. Wealthcan determine which choices are available to young men and women. Payingfor secondary school fees, for example, is often not feasible for poor house-holds. Wealth can also affect the relative valuation of different choices. Con-sider a woman’s decision to marry and leave the household. The higher thelevel of consumption she receives in her current household, the less attractiveany marital offer is. In addition, wealth can affect perceptions about oppor-tunities and well-being in the future. This might influence choices that in-volve trade-offs between current and future utilities. For instance, a young per-son’s decision about whether to engage in risky sex may be influenced by his orher present valuation of expected utility flows in the future.



Our data show that household wealth and the outcomes of interest for this


article are strongly correlated. In table 6 we report results for different out-comes measured in the 2009 round on respondents’ age and an indicator forhousehold wealth above the median in the baseline survey. Household wealthis defined by an asset index using principal component analysis. Wealth is as-sociated with a slower transition to adulthood. Relative to their poorer coun-terparts, young men and women of higher wealth are significantly more likelyto report attending school in the end line survey. They are also less likely to bemarried, to ever be pregnant, or to report ever having sex. The wealthier youngwomen are also less likely to report being sexually active.These trends observed in our data are consistent with findings in other

studies. In a similar setting in southern Malawi, cash transfers that increase in-come have been shown to significantly delay school exit, marriage, and fertilityonset of young women in Malawi and to reduce the prevalence of sexuallytransmitted diseases ðBaird, McIntosh, and Özler 2011; Baird et al. 2012Þ. Inaddition, several studies suggest that variation in income is linked to engagingin transactional sex—sexual relationships that are primarily motivated by ma-terial support to the female partner from theman ðe.g., Robinson andYeh 2011;Kohler and Thornton 2012; Burke, Gong, and Jones 2014Þ.

TABLE 6CORRELATION BETWEEN WEALTH AND TRANSITION BEHAVIOR

In School MarriedEver

PregnantNo. of

PregnanciesSexuallyActivea

MultiplePartnersb

Ever HadSex

Young Women

High wealth .254*** −.22*** −.134*** −.17*** −.14*** .003 −.13***ð.038Þ ð.041Þ ð.042Þ ð.052Þ ð.043Þ ð.011Þ ð.039Þ

Age −.06*** .032** .052*** .062*** .028** .002 .038***ð.012Þ ð.013Þ ð.013Þ ð.016Þ ð.013Þ ð.004Þ ð.012Þ

N 524 524 524 524 523 524 524R2 .114 .062 .044 .044 .025 .001 .035

Young Men

High wealth .119*** −.12*** −.156*** −.24*** −.082* −.027 −.06***ð.033Þ ð.040Þ ð.042Þ ð.054Þ ð.045Þ ð.030Þ ð.024Þ

Age −.05*** .048*** .055*** .063*** .028** −.011 .024***ð.009Þ ð.011Þ ð.012Þ ð.015Þ ð.012Þ ð.008Þ ð.007Þ

N 472 472 472 472 471 472 472R2 .095 .057 .071 .073 .018 .006 .041

Note. Mean outcomes as measured in the 2009 round. “High wealth” indicates that household wealthindex in 2007 is above median. Standard errors of the difference in means in parentheses.a Reports having sex in the 12 months before interview, excluding respondents who report being marriedin 2008.b Reports more than one sexual partner in the 12 months before interview.* Significant at 10% level.** Significant at 5% level.*** Significant at 1% level.



Given this background, we could expect that one’s response to VCTmay


differ depending on household wealth. To that end, we complement our studywith an ex post analysis of heterogeneous effects of testing by a household’swealth. The original randomization in the test offer was a simple randomiza-tion with no regard to wealth status at baseline.To explore heterogeneity in response to testing by a household’s wealth, we

perform the mean-comparison tests separately for individuals whose house-hold wealth index in the baseline is below or above the median. Finally, wealso explore heterogeneity in response to testing by both wealth and priorbeliefs. When splitting the sample by both wealth and prior belief categories,we note that there are smaller cell sizes and less power to detect impacts.In table 7, we present the effects of VCTon the two wealth groups. Overall,

we find negligible effects for young women. Among those from poorer house-holds who were attending school in 2008, the likelihood of attending schoolin 2009 increases by 19 percentage points after receiving a VCT offer. Thiseffect is statistically significant at the 10% level. The coefficients for the mar-ital and fertility outcomes are negative but not statistically significant. The ef-fects of the testing offer are not statistically significant on any of the outcomesfor young women of higher wealth.Table 8 presents the results of a test offer across four groups of young women:

women of lower wealth and some likelihood reported, women of lower wealthwho assign no likelihood to being infected, women of higher wealth who as-sign some likelihood, and women of higher wealth who assign no likelihood.Among women who are poorer and report some likelihood of being infected,the test offer reduces the likelihood of getting married by 30 percentage points.The likelihood of having ever been pregnant a year after the test is also reducedby 32 percentage points. Both of these effects are significant at the 5% level.Among the richer young women who assign some likelihood of being infected,one test out of six appears significant at the 10% level. Those who were neverpregnant by the time of the VCToffer were 20 percentage points more likely toever be pregnant a year later. The test offer does not affect these outcomes forthose who assign no likelihood to being infected, regardless of wealth status.Table 7 also presents results for the model interacting the test offer with a

household’s wealth for young men. We observe more impacts of testing forwealthier men, but not across all outcomes. The testing offer has statisticallysignificant effects on marital and fertility outcomes for young men of higherwealth. For these men, a test offer results in a 10 percentage point decrease inthe likelihood of getting married conditional on not being married at the timeof the test. In addition, the testing offer resulted in a decrease of 12 percent-age points in the likelihood of ever impregnating a sexual partner and a



TABLE 7EFFECT OF VOLUNTARY COUNSELING AND TESTING BY WEALTH

Stay inSchoola

GotMarriedb

FirstPregnancyc

No. ofPregnancies

SexuallyActived

MultiplePartnerse

Young Women

Low wealth:Control .457 .326 .475 .590 .477 .008Test offer .644 .265 .400 .518 .461 .022Difference −.187* .061 .075 .072 .016 −.014

ð.111Þ ð.067Þ ð.068Þ ð.078Þ ð.073Þ ð.015ÞHigh wealth:

Control .724 .155 .280 .410 .340 .009Test offer .767 .138 .301 .394 .442 .026Difference −.043 .016 −.022 .016 −.102 −.017


Low wealth:Control 35 86 101 122 86 122Test offer 45 102 115 137 102 137

High wealth:Control 58 97 93 110 97 110Test offer 73 130 136 155 129 155

Young Men

Low wealth:Control .765 .196 .337 .509 .588 .155Test offer .733 .175 .320 .534 .563 .107Difference .031 .021 .017 −.026 .026 .048

ð.158Þ ð.055Þ ð.067Þ ð.088Þ ð.071Þ ð.043ÞHigh wealth:

Control .609 .167 .242 .358 .556 .119Test offer .735 .064 .124 .216 .486 .086Difference −.127 .102** .118** .142** .060 .033


Low wealth:Control 17 102 101 116 102 116Test offer 15 97 97 131 96 131

High wealth:Control 23 90 91 109 90 109Test offer 34 109 105 116 109 116

Note. Mean outcomes as measured in the 2009 round. “High ðlowÞ wealth” indicates that householdwealth index in 2007 is above ðbelowÞ median. Standard errors of the difference in means in parentheses.a Excluding respondents in the fourth year of secondary school.b Conditional on not being married in the 2008 round.c Conditional on no pregnancies by the 2008 round.d Reports having sex in the 12 months before interview, excluding respondents who report being marriedin 2008.e Reports more than one sexual partner in the 12 months before interview.* Significant at 10% level.** Significant at 5% level.

680



TABLE 8EFFECT OF VOLUNTARY COUNSELING AND TESTING BY PRIOR BELIEFS AND WEALTH, YOUNG WOMEN

Stay inSchoola

GotMarriedb

FirstPregnancyc

No. ofPregnancies

SexuallyActived

MultiplePartnerse

Low wealth/likelihood:Control .667 .471 .708 .833 .647 0Test offer .667 .172 .387 .61 .414 0Difference 0 .299** .321** .223 .233 0

ð.327Þ ð.133Þ ð.131Þ ð.147Þ ð.152Þ 0High wealth/likelihood:

Control .714 .194 .267 .389 .484 0Test offer .6 .156 .471 .595 .613 .024Difference .114 .037 −.204* −.206 −.129 −.024

ð.182Þ ð.097Þ ð.120Þ ð.135Þ ð.127Þ ð.026ÞLow wealth/no likelihood:

Control .438 .29 .403 .511 .435 .011Test offer .636 .301 .405 .479 .479 .031Difference −.198 −.011 −.002 .032 −.044 −.02

ð.123Þ ð.077Þ ð.078Þ ð.091Þ ð.084Þ ð.021ÞHigh wealth/no likelihood:

Control .727 .138 .286 .411 .277 .014Test offer .81 .133 .245 .319 .388 .027Difference −0.083 .006 .041 .92 −.111 −.013


Low wealth/likelihood:Control 3 17 24 30 17 30Test offer 12 29 31 41 29 41

High wealth/no likelihood:Control 14 31 30 36 31 36Test offer 15 32 34 42 31 42

Low wealth/no likelihood:Control 32 69 77 92 69 92Test offer 33 73 84 96 73 96

High wealth/no likelihood:Control 44 65 63 73 65 73Test offer 58 98 102 113 98 113

Note. Mean outcomes as measured in the 2009 round. “Likelihood” represents whether respondentassigned any likelihood to being infected with HIV in the 2008 round. “High ðlowÞ wealth” indicates thathousehold wealth index in 2007 is above ðbelowÞ median. Standard errors of the difference in means inparentheses.a Excluding respondents in the fourth year of secondary school.b Conditional on not being married in the 2008 round.c Conditional on no pregnancies by the 2008 round.d Reports having sex in the 12 months before interview, excluding respondents who report being marriedin 2008.e Reports more than one sexual partner in the 12 months before interview.* Significant at 10% level.** Significant at 5% level.

681



reduction of 0.14 in the total number of pregnancies. Additional analysis of


wealth quartiles ðmore flexibility but smaller cellsÞ shows that these impactsare concentrated among the wealthiest quartile. The VCT offer did not re-sult in a statistically significant effect on any outcome for the poorer youngmen—results are both statistically insignificant and small in size.

DiscussionThe international community concerned with the AIDS epidemic in Africaviews HIV testing as a critical policy prescription needed to combat the dis-ease. This view has led to a huge investment in making HIV tests widely ac-cessible in most African countries. Testing people is a critical entry point intotreatment, which can lower infectiousness and, thus, serve to mitigate the spreadof the disease. But testing and the counseling that accompanies it are also sup-ported as a means to affect behaviors to reduce transmission rates. It is this latterrelationship we study here. We explore the response to HIV testing on sexualbehavior and the timing of important life events among young people in Ma-lawi. By looking at a random sample of men and women, we contribute to asmall but growing body of studies that randomize testing and explore behaviorchange in response to VCT.We find little response to an HIV test among our outcomes. We see no im-

pact of the VCT intervention on any of the behaviors by young women. Theresults for men suggest some slowdown in the transition toward adulthood inresponse to the VCT intervention as measured by marriage and impregnating asexual partner. Despite a generalized epidemic, these results are consistent withthe high rates of prior testing, the low rates of infectivity among young adults,and the low levels of reported likelihood of infection—characteristics of manysettings in sub-Saharan Africa where testing is touted as ameans to change behav-iors. The majority of young men and women report there is no or little chancethey are infected. As such, a test does not offer new information, although it doesprovide confirmation about one’s prior beliefs.We do look for heterogeneity in response to a test through a set of ex post

analysis looking at prior beliefs andwealth. These are ex post in the sense that theoriginal study was a randomization without regard to these traits. The wealtheffects differ for women and men. The poorer young women, who assignedsome likelihood to being infected before receiving theVCToffer, are less likely totransition into marriage or fertility a year after the intervention. Among theyoung men, it is the wealthier ones who are less likely to be married or to fatherchildren a year after the VCToffer. This heterogeneity by socioeconomic statuscan result from the different set of opportunities richer and poorer young adultsface and the different expectations they have about their futures. It could also be



that testing influences decisions similarly, but because of the different propen-


sities to engage in the different behaviors, independent of testing, we can onlyobserve significant changes in behaviors for some groups.

ReferencesAngotti, Nicole. 2010. “Working Outside the Box: How HIV Counselors in Sub-SaharanAfrica AdaptWesternTestingNorms.” Social Science andMedicine 71, no. 5:986–93.

Angotti, Nicole, Kim Yi Dionne, and Lauren Gaydosh. 2011. “An Offer You Can’tRefuse? Provider-Initiated HIV Testing in Antenatal Clinics in Rural Malawi.”Health Policy and Planning 26, no. 4:307–15.

Baird, Sarah, Erick Gong, Craig McIntosh, and Berk Özler. 2014. “The Heteroge-neous Effects of HIV Testing.” Journal of Health Economics 37:98–112.

Baird, Sarah, Craig McIntosh, and Berk Özler. 2011. “Cash or Condition? EvidencefromaCashTransferExperiment.”Quarterly Journal ofEconomics126, no. 4:1709–53.

Baird, Sarah J., Richard S. Garfein, Craig T. McIntosh, and Berk Özler. 2012. “Effectof a Cash Transfer Programme for Schooling on Prevalence of HIV and HerpesSimplex Type 2 in Malawi: A Cluster Randomised Trial.” Lancet 379, no. 9823:1320–29.

Beegle, Kathleen, and Michelle Poulin. 2013. “Migration and the Transition to Adult-hood in Contemporary Malawi.” Annals of the American Academy of Political andSocial Science 648, no. 1:38–51.

Boozer, Michael, and Tomas Philipson. 2000. “The Impact of Public Testing forHuman Immunodeficiency Virus.” Journal of Human Resources 35, no. 3:419–46.

Burke, Marshall, Erick Gong, and Kelly Jones. 2014. “Income Shocks and HIV inAfrica.” Economic Journal ðforthcomingÞ.

Clark, Shelley, Michelle Poulin, and Hans-Peter Kohler. 2009. “Marriage Aspirationsand HIV/AIDS in Rural Malawi.” Journal of Marriage and the Family 71:396–416.

Delavande, Adeline, and Hans-Peter Kohler. 2012. “The Impact of HIV Testing onSubjective Expectations and Risky Behaviors in Malawi.” Demography 49, no. 3:1011–36.

Gersovitz, Mark. 2011. “HIV Testing: Principles and Practice.” World Bank ResearchObserver 26, no. 1:1–41.

Glynn, Judith R., Michel Caraël, Anne Buvé, Rosemary M. Musonda, Maina Ka-hindo, and Study Group on the Heterogeneity of HIV Epidemics in African Cities.2003. “HIV Risk in Relation to Marriage in Areas with High Prevalence of HIVInfection.” Journal of Acquired Immune Deficiency Syndromes 33, no. 4:526–35.

Gong, Erick. 2015. “HIV Testing and Risky Sexual Behavior.” Economic Journal 125,no. 582:32–60.

Government of Malawi. 2003. “National HIV/AIDS Policy: A Call to Action.”Government of Malawi, Lilongwe.

Kohler, Hans-Peter, and Rebecca L. Thornton. 2012. “Conditional Cash Transfersand HIV/AIDS Prevention: Unconditionally Promising?”World Bank Economic Re-view 26, no. 2:165–90.



Magruder, Jeremy. 2011. “Marital Shopping and Epidemic AIDS.” Demography 14,no. 4:1401–28.


NSO ðNational Statistical OfficeÞ and ICF Macro. 2011. Malawi Demographic andHealth Survey, 2010. Zomba and Calverton, MD: NSO and ICF Macro.

Potts, Malcolm, Daniel T. Halperin, Douglas Kirby, Ann Swidler, Elliot Marseille,Jeffrey D. Klausner, Norman Hearst, Richard G. Wamai, James G. Kahn, and JuliaWalsh. 2008. “Reassessing HIV Prevention.” Science 320:749–50.

Poulin, Michelle. 2007. “Sex, Money, and Premarital Partnerships in SouthernMalawi.” Social Science and Medicine 65, no. 11:2383–93.

Poulin, Michelle, and Kathleen Beegle. 2014. “Out-of-Wedlock Fertility and theTiming of First Marriage in Malawi.” Photocopy, World Bank, Washington, DC.

Poulin, Michelle, Kate Dovel, and Susan Watkins. 2014. “AIDS: Wealthy Men andthe ‘Vulnerable Women’ Client Category.” Photocopy, World Bank, Washington,DC.

Robinson, James, and Ethan Yeh. 2011. “Transactional Sex as a Response to Risk inWestern Kenya.” American Economic Journal: Applied Economics 3, no. 1:35–64.

Smith, Kirsten P., and Susan Cotts Watkins. 2005. “Perceptions of Risk and Strategiesfor Prevention: Responses to HIV/AIDS in Rural Malawi.” Social Science andMedicine 60, no. 3:649–60.

Thornton, Rebecca. 2008. “The Demand for and Impact of Learning HIV Status:Evidence from a Field Experiment.” American Economics Review 98, no. 5:1829–63.

Trinitapoli, Jenny, and Sara Yeatman. 2011. “Uncertainty and Fertility in a Gener-alized AIDS Epidemic.” American Sociological Review 76, no. 6:935–54.

Watkins, Susan. 2004. “Navigating the AIDS Epidemic in Rural Malawi.” Populationand Development Review 30, no. 4:673–705.



HIV testing, behavior change, and the transition to adulthood in Malawi

Documents